wiki:PyYAMLDocumentation

Version 23 (modified by xi, 9 years ago) (diff)

--

PyYAML Documentation

work in progress

RPG-ish descriptions are stolen from  the Angband rogue-like game. Names of the heroes are generated with  MudNames.

This documentation is very brief and incomplete. Feel free to fix or improve it.

Introduction

Start with importing the yaml package.

>>> import yaml

Loading YAML

Warning: It is not safe to call yaml.load with any data received from an untrusted source! yaml.load is as powerful as pickle.load and so may call any Python function. Check the yaml.safe_load function though.

The function yaml.load converts a YAML document to a Python object.

>>> yaml.load("""
... - Hesperiidae
... - Papilionidae
... - Apatelodidae
... - Epiplemidae
... """)

['Hesperiidae', 'Papilionidae', 'Apatelodidae', 'Epiplemidae']

yaml.load accepts a string, a Unicode string, an open file object, or an open Unicode file object. A string or a file must be encoded with utf-8, utf-16-be or utf-16-le encoding. yaml.load detects the encoding by checking the BOM (byte order mark) sequence at the beginning of the string/file. If no BOM is present, the utf-8 encoding is assumed.

yaml.load returns a Python object.

>>> yaml.load(u"""
... hello: Привет!
... """)

{'hello': u'\u041f\u0440\u0438\u0432\u0435\u0442!'}

>>> stream = file('document.yaml', 'r')    # 'document.yaml' contains a single YAML document.
>>> yaml.load(stream)
[...]    # A Python object corresponding to the document.

If a string or a file contains several documents, you may load them all with the yaml.load_all function.

>>> documents = """
... ---
... name: The Set of Gauntlets 'Pauraegen'
... description: >
...     A set of handgear with sparks that crackle
...     across its knuckleguards.
... ---
... name: The Set of Gauntlets 'Paurnen'
... description: >
...   A set of gauntlets that gives off a foul,
...   acrid odour yet remains untarnished.
... ---
... name: The Set of Gauntlets 'Paurnimmen'
... description: >
...   A set of handgear, freezing with unnatural cold.
... """

>>> for data in yaml.load_all(documents):
...     print data

{'description': 'A set of handgear with sparks that crackle across its knuckleguards.\n',
'name': "The Set of Gauntlets 'Pauraegen'"}
{'description': 'A set of gauntlets that gives off a foul, acrid odour yet remains untarnished.\n',
'name': "The Set of Gauntlets 'Paurnen'"}
{'description': 'A set of handgear, freezing with unnatural cold.\n',
'name': "The Set of Gauntlets 'Paurnimmen'"}

PyYAML allows you to construct a Python object of any type.

>>> yaml.load("""
... none: [~, null]
... bool: [true, false, on, off]
... int: 42
... float: 3.14159
... list: [LITE, RES_ACID, SUS_DEXT]
... dict: {hp: 13, sp: 5}
... """)

{'none': [None, None], 'int': 42, 'float': 3.1415899999999999,
'list': ['LITE', 'RES_ACID', 'SUS_DEXT'], 'dict': {'hp': 13, 'sp': 5},
'bool': [True, False, True, False]}

Even instances of Python classes can be constructed using the !!python/object tag.

>>> class Hero:
...     def __init__(self, name, hp, sp):
...         self.name = name
...         self.hp = hp
...         self.sp = sp
...     def __repr__(self):
...         return "%s(name=%r, hp=%r, sp=%r)" % (
...             self.__class__.__name__, self.name, self.hp, self.sp)

>>> yaml.load("""
... !!python/object:__main__.Hero
... name: Welthyr Syxgon
... hp: 1200
... sp: 0
... """)

Hero(name='Welthyr Syxgon', hp=1200, sp=0)

Note that the ability to construct an arbitrary Python object may be dangerous if you receive a YAML document from an untrusted source such as Internet. The function yaml.safe_load limits this ability to simple Python objects like integers or lists.

Dumping YAML

The yaml.dump function accepts a Python object and produces a YAML document.

>>> print yaml.dump({'name': 'Silenthand Olleander', 'race': 'Human',
... 'traits': ['ONE_HAND', 'ONE_EYE']})

name: Silenthand Olleander
race: Human
traits: [ONE_HAND, ONE_EYE]

yaml.dump accepts the second optional argument, which must be an open file. In this case, yaml.dump will write the produced YAML document into the file. Otherwise, yaml.dump returns the produced document.

>>> stream = file('document.yaml', 'w')
>>> yaml.dump(data, stream)    # Write a YAML representation of data to 'document.yaml'.
>>> print yaml.dump(data)      # Output the document to the screen.

If you need to dump several YAML documents to a single stream, use the function yaml.dump_all. yaml.dump_all accepts a list or a generator producing Python objects to be serialized into a YAML document. The second optional argument is an open file.

>>> print yaml.dump([1,2,3], explicit_start=True)
--- [1, 2, 3]

>>> print yaml.dump_all([1,2,3], explicit_start=True)
--- 1
--- 2
--- 3

You may even dump instances of Python classes.

>>> class Hero:
...     def __init__(self, name, hp, sp):
...         self.name = name
...         self.hp = hp
...         self.sp = sp
...     def __repr__(self):
...         return "%s(name=%r, hp=%r, sp=%r)" % (
...             self.__class__.__name__, self.name, self.hp, self.sp)

>>> print yaml.dump(("Galain Ysseleg", hp=-3, sp=2))

!!python/object:__main__.Hero {hp: -3, name: Galain Ysseleg, sp: 2}

yaml.dump supports a number of keyword arguments that specify formatting details for the emitter. For instance, you may set the preferred intendation and width, use the canonical YAML format or force preferred style for scalars and collections.

>>> print yaml.dump(range(50))
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22,
  23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
  43, 44, 45, 46, 47, 48, 49]

>>> print yaml.dump(range(50), width=50, indent=4)
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
    16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
    28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39,
    40, 41, 42, 43, 44, 45, 46, 47, 48, 49]

>>> print yaml.dump(range(5), canonical=True)
---
!!seq [
  !!int "0",
  !!int "1",
  !!int "2",
  !!int "3",
  !!int "4",
]

>>> print yaml.dump(range(5), default_flow_style=False)
- 0
- 1
- 2
- 3
- 4

>>> print yaml.dump(range(5), default_flow_style=True, default_style='"')
[!!int "0", !!int "1", !!int "2", !!int "3", !!int "4"]

Constructors, representers, resolvers

You may define your own application-specific tags. The easiest way to do it is to define a subclass of yaml.YAMLObject:

>>> yaml.load("""
... --- !Monster
... name: Cave spider
... hp: [2,6]    # 2d6
... ac: 16
... attacks: [BITE, HURT]
... """)

Monster(name='Cave spider', hp=[2, 6], ac=16, attacks=['BITE', 'HURT'])

>>> print yaml.dump(Monster(
...     name='Cave lizard', hp=[3,6], ac=16, attacks=['BITE','HURT']))

!Monster
ac: 16
attacks: [BITE, HURT]
hp: [3, 6]
name: Cave lizard

yaml.YAMLObject uses metaclass magic to register a constructor, which transforms a YAML node to a class instance, and a representer, which serializes a class instance to a YAML node.

If you don't want to use metaclasses, you may register your constructors and representers using the functions yaml.add_constructor and yaml.add_representer. For instance, you may want to add a constructor and a representer for the following Dice class:

>>> class Dice(tuple):
...     def __new__(cls, a, b):
...         return tuple.__new__(cls, [a, b])
...     def __repr__(self):
...         return "%sd%s" % self

>>> print Dice(3,6)
3d6

First we define a representer, that convert a dice object to scalar node with the tag !dice and register it.

>>> def dice_representer(dumper, data):
...     return dumper.represent_scalar(u'!dice', unicode(data))

>>> yaml.add_representer(Dice, dice_representer)

Now you may dump an instance of the Dice object:

>>> print yaml.dump({'gold': Dice(10,6)})
{gold: !dice '10d6'}

Let us add the code to construct a Dice object:

>>> def dice_constructor(loader, node):
...     value = loader.construct_scalar(node)
...     a, b = map(int, value.split('d'))
...     return Dice(a, b)

>>> yaml.add_constructor(u'!dice', dice_constructor)

Then you may load a Dice object as well:

>>> print yaml.load("""
... initial hit points: !dice 8d4
... """)

{'initial hit points': 8d4

You might want to not specify the tag !dice everywhere. There is a way to teach PyYAML that any untagged plain scalar that looks like XdY has the implicit tag !dice. Use add_implicit_resolver:

>>> import re
>>> pattern = re.compile(r'^\d+d\d+$')
>>> yaml.add_implicit_resolver(u'!dice', pattern)

Now you don't have to specify the tag to define a Dice object:

>>> print yaml.dump({'treasure': Dice(10,20)})

{treasure: 10d20}

>>> print yaml.load("""
... damage: 5d10
... """)

{'damage': 5d10}

YAML syntax

A good introduction to the YAML syntax is  Chapter 2 of the YAML specification.

You may also check  the YAML cookbook. Note that it is focused on a Ruby implementation and uses the old YAML 1.0 syntax.

Here we present most common YAML constructs together with the corresponding Python objects.

Documents

YAML stream is a collection of zero or more documents. An empty stream contains no documents. Documents are separated with ---. Documents may optionally end with .... A single document may or may not be marked with ---.

Example of an implicit document:

- Multimedia
- Internet
- Education

Example of an explicit document:

---
- Afterstep
- CTWM
- Oroborus
...

Example of several documents in the same stream:

---
- Ada
- APL
- ASP
- Assembly
- Awk
---
- Basic
---
- C
- C#    # Note that comments are denoted with ' #' (space and #).
- C++
- Cold Fusion

Block sequences

In the block context, sequence entries are denoted by - (dash and space):

# YAML
- The Dagger 'Narthanc'
- The Dagger 'Nimthanc'
- The Dagger 'Dethanc'
# Python
["The Dagger 'Narthanc'", "The Dagger 'Nimthanc'", "The Dagger 'Dethanc'"]

Block sequences can be nested:

# YAML
-
  - HTML
  - LaTeX
  - SGML
  - VRML
  - XML
  - YAML
-
  - BSD
  - GNU Hurd
  - Linux
# Python
[['HTML', 'LaTeX', 'SGML', 'VRML', 'XML', 'YAML'], ['BSD', 'GNU Hurd', 'Linux']]

It's not necessary to start a nested sequence with a new line:

# YAML
- 1.1
- - 2.1
  - 2.2
- - - 3.1
    - 3.2
    - 3.3
# Python
[1.1, [2.1, 2.2], [[3.1, 3.2, 3.3]]]

A block sequence may be nested to a block mapping. Note that in this case it is not necessary to indent the sequence.

# YAML
left hand:
- Ring of Teleportation
- Ring of Speed
right hand:
- Ring of Resist Fire
- Ring of Resist Cold
- Ring of Resist Poison
# Python
{'right hand': ['Ring of Resist Fire', 'Ring of Resist Cold', 'Ring of Resist Poison'],
'left hand': ['Ring of Teleportation', 'Ring of Speed']}

Block mappings

In the block context, keys and values of mappings are separated by : (colon and space):

# YAML
base armor class: 0
base damage: [4,4]
plus to-hit: 12
plus to-dam: 16
plus to-ac: 0
# Python
{'plus to-hit': 12, 'base damage': [4, 4], 'base armor class': 0, 'plus to-ac': 0, 'plus to-dam': 16}

Complex keys are denoted with ? (question mark and space):

# YAML
? !!python/tuple [0,0]
: The Hero
? !!python/tuple [0,1]
: Treasure
? !!python/tuple [1,0]
: Treasure
? !!python/tuple [1,1]
: The Dragon
# Python
{(0, 1): 'Treasure', (1, 0): 'Treasure', (0, 0): 'The Hero', (1, 1): 'The Dragon'}

Block mapping can be nested:

# YAML
hero:
  hp: 34
  sp: 8
  level: 4
orc:
  hp: 12
  sp: 0
  level: 2
# Python
{'hero': {'hp': 34, 'sp': 8, 'level': 4}, 'orc': {'hp': 12, 'sp': 0, 'level': 2}}

A block mapping may be nested in a block sequence:

# YAML
- name: PyYAML
  status: 4
  license: MIT
  language: Python
- name: PySyck
  status: 5
  license: BSD
  language: Python
# Python
[{'status': 4, 'language': 'Python', 'name': 'PyYAML', 'license': 'MIT'},
{'status': 5, 'license': 'BSD', 'name': 'PySyck', 'language': 'Python'}]

Flow collections

The syntax of flow collections in YAML is very close to the syntax of list and dictionary constructors in Python:

# YAML
{ str: [15, 17], con: [16, 16], dex: [17, 18], wis: [16, 16], int: [10, 13], chr: [5, 8] }
# Python
{'dex': [17, 18], 'int': [10, 13], 'chr': [5, 8], 'wis': [16, 16], 'str': [15, 17], 'con': [16, 16]}

Scalars

There are 5 styles of scalars in YAML: plain, single-quoted, double-quoted, literal, and folded:

# YAML
plain: Scroll of Remove Curse
single-quoted: 'EASY_KNOW'
double-quoted: "?"
literal: |    # Borrowed from http://www.kersbergen.com/flump/religion.html
  by hjw              ___
     __              /.-.\
    /  )_____________\\  Y
   /_ /=== == === === =\ _\_
  ( /)=== == === === == Y   \
   `-------------------(  o  )
                        \___/
folded: >
  It removes all ordinary curses from all equipped items.
  Heavy or permanent curses are unaffected.
# Python
{'folded': 'It removes all ordinary curses from all equipped items. \
Heavy or permanent curses are unaffected.\n',
'literal': 'by hjw              ___\n   __              /.-.    /  )_____________\\  Y\n\
 /_ /=== == === === =\\ _\\_\n( /)=== == === === == Y      `-------------------(  o  )\n\
                      \\___/\n',
'single-quoted': 'EASY_KNOW', 'double-quoted': '?', 'plain': 'Scroll of Remove Curse'}

Each style has its own quirks. A plain scalar does not use indicators to denote its start and end, therefore it's the most restricted style. Its natural applications are names of attributes and parameters.

Using single-quoted scalars, you may express any value that does not contain special characters. No escaping occurs for single quoted scalars except that duplicate quotes '' are replaced with a single quote '.

Double-quoted is the most powerful style and the only style that can express any scalar value. Double-quoted scalars allow escaping. Using escaping sequences \x** and \u****, you may express any ASCII or Unicode character.

There are two kind of block scalar styles: literal and folded. The literal style is the most suitable style for large block of text such as source code. The folded style is similar to the literal style, but two consequent non-empty lines are joined to a single line separated by a space character.

Aliases

Note that PyYAML does not yet support recursive objects.

Using YAML you may represent objects of arbitrary graph-like structures. If you want to refer to the same object from different parts of a document, you need to use anchors and aliases.

Anchors are denoted by the & indicator while aliases are denoted by *. For instance, the document

left hand: &A The Bastard Sword of Eowyn
right hand: *A

expresses the idea of a hero holding a heavy sword in both hands.

Tags

Tags are used to denote the type of a YAML node. Standard YAML tags are defined at  http://yaml.org/type/index.html.

Tags may be implicit:

boolean: true
integer: 3
float: 3.14
{'boolean': True, 'integer': 3, 'float': 3.14}

or explicit:

boolean: !!bool "true"
integer: !!int "3"
float: !!float "3.14"
{'boolean': True, 'integer': 3, 'float': 3.14}

Plain scalars without explicitly defined tag are subject to implicit tag resolution. The scalar value is checked against a set of regular expressions and if one of them matches, the corresponding tag is assigned to the scalar. PyYAML allows an application to add custom implicit tag resolvers.

YAML tags and Python types

The following table describes how nodes with different tags are converted to Python objects.

YAML tag Python type
Standard YAML tags
!!null None
!!bool bool
!!int int or long
!!float float
!!binary str
!!timestamp datetime.datetime
!!omap, !!pairs list of pairs
!!set set
!!str str or unicode
!!seq list
!!map dict
Python-specific tags
!!python/none None
!!python/bool bool
!!python/str str
!!python/unicode unicode
!!python/int int
!!python/long long
!!python/float float
!!python/complex complex
!!python/list list
!!python/tuple tuple
!!python/dict dict
Complex Python tags
!!python/name:module.name module.name
!!python/module:package.module package.module
!!python/object:module.cls module.cls instance
!!python/object/new:module.cls module.cls instance
!!python/object/apply:module.f value of f(...)

String conversion

There are four tags that are converted to str and unicode values: !!str, !!binary, !!python/str, and !!python/unicode.

!!str-tagged scalars are converted to str objects if its value is ASCII. Otherwise it is converted to unicode. !!binary-tagged scalars are converted to str objects with its value decoded using the base64 encoding. !!python/str scalars are converted to str objects encoded with utf-8 encoding. !!python/unicode scalars are converted to unicode objects.

Conversely, a str object is converted to

  1. a !!str scalar if its value is ASCII.
  2. a !!python/str scalar if its value is a correct utf-8 sequence.
  3. a !!binary scalar otherwise.

A unicode object is converted to

  1. a !!python/unicode scalar if its value is ASCII.
  2. a !!str scalar otherwise.

Names and modules

In order to represent static Python objects like functions or classes, you need to use a complex !!python/name tag. For instance, the function yaml.dump can be represented as

!!python/name:yaml.dump

Similarly, modules are represented using the tag !python/module:

!!python/module:yaml

Objects

Any pickleable object can be serialized using the !!python/object tag:

!!python/object:module.Class { attribute: value, ... }

In order to support the pickle protocol, two additional forms of the !!python/object tag are provided:

!!python/object/new:module.Class
args: [argument, ...]
kwds: {key: value, ...}
state: ...
listitems: [item, ...]
dictitems: [key: value, ...]
!!python/object/apply:module.function
args: [argument, ...]
kwds: {key: value, ...}
state: ...
listitems: [item, ...]
dictitems: [key: value, ...]

If only the args field is non-empty, the above records can be shortened:

!!python/object/new:module.Class [argument, ...]
!!python/object/apply:module.function [argument, ...]

Reference

Warning: API stability is not guaranteed'''

The yaml package

scan(stream, Loader=Loader)

scan(stream) scans the given stream and produces a sequence of tokens.

parse(stream, Loader=Loader)

emit(events, stream=None, Dumper=Dumper,
    canonical=None,
    indent=None,
    width=None,
    allow_unicode=None,
    line_break=None)

parse(stream) parses the given stream and produces a sequence of parsing events.

emit(events, stream=None) serializes the given sequence of parsing events and write them to the stream. if stream is None, it returns the produced stream.

compose(stream, Loader=Loader)
compose_all(stream, Loader=Loader)

serialize(node, stream=None, Dumper=Dumper,
    encoding='utf-8',
    explicit_start=None,
    explicit_end=None,
    version=None,
    tags=None,
    canonical=None,
    indent=None,
    width=None,
    allow_unicode=None,
    line_break=None)
serialize_all(nodes, stream=None, Dumper=Dumper, ...)

compose(stream) parses the given stream and returns the root of the representation graph for the first document in the stream. If there are no documents in the stream, it returns None.

compose_all(stream) parses the given stream and returns a sequence of representation graphs corresponding to the documents in the stream.

serialize(node, stream=None) serializes the given representation graph into the stream. If stream is None, it returns the produced stream.

serialize_all(node, stream=None) serializes the given sequence of representation graphs into the given stream. If stream is None, it returns the produced stream.

load(stream, Loader=Loader)
load_all(stream, Loader=Loader)

safe_load(stream)
safe_load_all(stream)

dump(data, stream=None, Dumper=Dumper,
    default_style=None,
    default_flow_style=None,
    encoding='utf-8',
    explicit_start=None,
    explicit_end=None,
    version=None,
    tags=None,
    canonical=None,
    indent=None,
    width=None,
    allow_unicode=None,
    line_break=None)
dump_all(data, stream=None, Dumper=Dumper, ...)

safe_dump(data, stream=None, ...)
safe_dump_all(data, stream=None, ...)

load(stream) parses the given stream and returns a Python object constructed from for the first document in the stream. If there are no documents in the stream, it returns None.

load_all(stream) parses the given stream and returns a sequence of Python objects corresponding to the documents in the stream.

safe_load(stream) parses the given stream and returns a Python object constructed from for the first document in the stream. If there are no documents in the stream, it returns None. safe_load recognizes only standard YAML tags and cannot construct an arbitrary Python object.

safe_load_all(stream) parses the given stream and returns a sequence of Python objects corresponding to the documents in the stream. safe_load_all recognizes only standard YAML tags and cannot construct an arbitrary Python object.

dump(data, stream=None) serializes the given Python object into the stream. If stream is None, it returns the produced stream.

dump_all(data, stream=None) serializes the given sequence of Python objects into the given stream. If stream is None, it returns the produced stream. Each object is represented as a YAML document.

safe_dump(data, stream=None) serializes the given Python object into the stream. If stream is None, it returns the produced stream. safe_dump produces only standard YAML tags and cannot represent an arbitrary Python object.

safe_dump_all(data, stream=None) serializes the given sequence of Python objects into the given stream. If stream is None, it returns the produced stream. Each object is represented as a YAML document. safe_dump_all produces only standard YAML tags and cannot represent an arbitrary Python object.

def constructor(loader, node):
    # ...
    return data

def multi_constructor(loader, tag_suffix, node):
    # ...
    return data

add_constructor(tag, constructor, Loader=Loader)
add_multi_constructor(tag_prefix, multi_constructor, Loader=Loader)

add_constructor(tag, constructor) allows to specify a constructor for the given tag. A constructor is a function that converts a node of a YAML representation graph to a native Python object. A constructor accepts an instance of Loader and a node and returns a Python object.

add_multi_constructor(tag_prefix, multi_constructor) allows to specify a multi_constructor for the given tag_prefix. A multi-constructor is a function that converts a node of a YAML representation graph to a native Python object. A multi-constructor accepts an instance of Loader, the suffix of the node tag, and a node and returns a Python object.

def representer(dumper, data):
    # ...
    return node

def multi_representer(dumper, data):
    # ...
    return node

add_representer(data_type, representer, Dumper=Dumper)
add_multi_representer(base_data_type, multi_representer, Dumper=Dumper)

add_representer(data_type, representer) allows to specify a representer for Python objects of the given data_type. A representer is a function that converts a native Python object to a node of a YAML representation graph. A representer accepts an instance of Dumper and an object and returns a node.

add_multi_representer(base_data_type, multi_representer) allows to specify a multi_representer for Python objects of the given base_data_type or any of its subclasses. A multi-representer is a function that converts a native Python object to a node of a YAML representation graph. A multi-representer accepts an instance of Dumper and an object and returns a node.

add_implicit_resolver(tag, regexp, first, Loader=Loader, Dumper=Dumper)
add_path_resolver(tag, path, kind, Loader=Loader, Dumper=Dumper)

add_implicit_resolver(tag, regexp, first) adds an implicit tag resolver for plain scalars. If the scalar value is matched the given regexp, it is assigned the tag. first is a list of possible initial characters or None.

add_path_resolver(tag, path, kind) adds a path-based implicit tag resolver. A path is a list of keys that form a path to a node in the representation graph. Paths elements can be string values, integers, or None. The kind of a node can be str, list, dict, or None.

YAMLError

YAMLError()

If YAML parser encounters an error condition, it raises an exception which is an instance of YAMLError or of its subclass. An application may catch this exception and warn a user.

try:
    config = yaml.load(file('config.yaml', 'r'))
except yaml.YAMLError, exc:
    print "Error in configuration file:", exc

Mark

Mark(name, index, line, column, buffer, pointer)

An instance of Mark points to a certain position in the input stream. name is the name of the stream, for instance it may be the filename if the input stream is a file. line and column is the line and column of the position (starting from 0). buffer, when it is not None, is a part of the input stream that contain the position and pointer refers to the position in the buffer.

Tokens

Tokens are produced by a YAML scanner. They are not really useful except for low-level YAML applications such as syntax highlighting.

The PyYAML scanner produces the following types of tokens:

StreamStartToken(encoding, start_mark, end_mark) # Start of the stream.
StreamEndToken(start_mark, end_mark) # End of the stream.
DirectiveToken(name, value, start_mark, end_mark) # YAML directive, either %YAML or %TAG.
DocumentStartToken(start_mark, end_mark) # '---'.
DocumentEndToken(start_mark, end_mark) # '...'.
BlockSequenceStartToken(start_mark, end_mark) # Start of a new block sequence.
BlockMappingStartToken(start_mark, end_mark) # Start of a new block mapping.
BlockEndToken(start_mark, end_mark) # End of a block collection.
FlowSequenceStartToken(start_mark, end_mark) # '['.
FlowMappingStartToken(start_mark, end_mark) # '{'.
FlowSequenceEndToken(start_mark, end_mark) # ']'.
FlowMappingEndToken(start_mark, end_mark) # '}'.
KeyToken(start_mark, end_mark) # Either '?' or start of a simple key.
ValueToken(start_mark, end_mark) # ':'.
BlockEntryToken(start_mark, end_mark) # '-'.
FlowEntryToken(start_mark, end_mark) # ','.
AliasToken(value, start_mark, end_mark) # '*value'.
AnchorToken(value, start_mark, end_mark) # '&value'.
TagToken(value, start_mark, end_mark) # '!value'.
ScalarToken(value, plain, style, start_mark, end_mark) # 'value'.

start_mark and end_mark denote the beginning and the and of a token.

Example:

>>> document = """
... ---
... block sequence:
... - BlockEntryToken
... block mapping:
...   ? KeyToken
...   : ValueToken
... flow sequence: [FlowEntryToken, FlowEntryToken]
... flow mapping: {KeyToken: ValueToken}
... anchors and tags:
... - &A !!int '5'
... - *A
... ...
... """

>>> for token in yaml.scan(document):
...     print token

StreamStartToken(encoding='utf-8')

DocumentStartToken()

BlockMappingStartToken()

KeyToken()
ScalarToken(plain=True, style=None, value=u'block sequence')

ValueToken()
BlockEntryToken()
ScalarToken(plain=True, style=None, value=u'BlockEntryToken')

KeyToken()
ScalarToken(plain=True, style=None, value=u'block mapping')

ValueToken()
BlockMappingStartToken()


KeyToken()
ScalarToken(plain=True, style=None, value=u'KeyToken')
ValueToken()
ScalarToken(plain=True, style=None, value=u'ValueToken')
BlockEndToken()

KeyToken()
ScalarToken(plain=True, style=None, value=u'flow sequence')

ValueToken()
FlowSequenceStartToken()
ScalarToken(plain=True, style=None, value=u'FlowEntryToken')
FlowEntryToken()
ScalarToken(plain=True, style=None, value=u'FlowEntryToken')
FlowSequenceEndToken()

KeyToken()
ScalarToken(plain=True, style=None, value=u'flow mapping')

ValueToken()
FlowMappingStartToken()
KeyToken()
ScalarToken(plain=True, style=None, value=u'KeyToken')
ValueToken()
ScalarToken(plain=True, style=None, value=u'ValueToken')
FlowMappingEndToken()

KeyToken()
ScalarToken(plain=True, style=None, value=u'anchors and tags')

ValueToken()
BlockEntryToken()
AnchorToken(value=u'A')
TagToken(value=(u'!!', u'int'))
ScalarToken(plain=False, style="'", value=u'5')
BlockEntryToken()
AliasToken(value=u'A')

BlockEndToken()

DocumentEndToken()

StreamEndToken()

Events

Events are used by the low-level Parser and Emitter interfaces, which are similar to the SAX API. While the Parser parses a YAML stream and produces a sequence of events, the Emitter accepts a sequence of events and emits a YAML stream.

The following events are defined:

StreamStartEvent(encoding, start_mark, end_mark)
StreamEndEvent(start_mark, end_mark)
DocumentStartEvent(explicit, version, tags, start_mark, end_mark)
DocumentEndEvent(start_mark, end_mark)
SequenceStartEvent(anchor, tag, implicit, flow_style, start_mark, end_mark)
SequenceEndEvent(start_mark, end_mark)
MappingStartEvent(anchor, tag, implicit, flow_style, start_mark, end_mark)
MappingEndEvent(start_mark, end_mark)
AliasEvent(anchor, start_mark, end_mark)
ScalarEvent(anchor, tag, implicit, value, style, start_mark, end_mark)

The flow_style flag indicates if a collection is block or flow. The possible values are None, True, False. The style flag of a scalar event indicates the style of the scalar. Possible values are None, '', '\'', '"', '|', '>'. The implicit flag of a collection start event indicates if the tag may be omitted when the collection is emitted. The implicit flag of a scalar event is a pair of boolean values that indicate if the tag may be omitted when the scalar is emitted in a plain and non-plain style correspondingly.

Example:

>>> document = """
... scalar: &A !!int '5'
... alias: *A
... sequence: [1, 2, 3]
... mapping: [1: one, 2: two, 3: three]
... """

>>> for event in yaml.parse(document):
...     print event

StreamStartEvent()

DocumentStartEvent()

MappingStartEvent(anchor=None, tag=None, implicit=True)

ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'scalar')
ScalarEvent(anchor=u'A', tag=u'tag:yaml.org,2002:int', implicit=(False, False), value=u'5')

ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'alias')
AliasEvent(anchor=u'A')

ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'sequence')
SequenceStartEvent(anchor=None, tag=None, implicit=True)
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'1')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'2')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'3')
SequenceEndEvent()

ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'mapping')
MappingStartEvent(anchor=None, tag=None, implicit=True)
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'1')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'one')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'2')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'two')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'3')
ScalarEvent(anchor=None, tag=None, implicit=(True, False), value=u'three')
MappingEndEvent()

MappingEndEvent()

DocumentEndEvent()

StreamEndEvent()

>>> print yaml.emit([
...     yaml.StreamStartEvent(encoding='utf-8'),
...     yaml.DocumentStartEvent(explicit=True),
...     yaml.MappingStartEvent(anchor=None, tag=u'tag:yaml.org,2002:map', implicit=True, flow_style=False),
...     yaml.ScalarEvent(anchor=None, tag=u'tag:yaml.org,2002:str', implicit=(True, True), value=u'agile languages'),
...     yaml.SequenceStartEvent(anchor=None, tag=u'tag:yaml.org,2002:seq', implicit=True, flow_style=True),
...     yaml.ScalarEvent(anchor=None, tag=u'tag:yaml.org,2002:str', implicit=(True, True), value=u'Python'),
...     yaml.ScalarEvent(anchor=None, tag=u'tag:yaml.org,2002:str', implicit=(True, True), value=u'Perl'),
...     yaml.ScalarEvent(anchor=None, tag=u'tag:yaml.org,2002:str', implicit=(True, True), value=u'Ruby'),
...     yaml.SequenceEndEvent(),
...     yaml.MappingEndEvent(),
...     yaml.DocumentEndEvent(explicit=True),
...     yaml.StreamEndEvent(),
... ])

---
agile languages: [Python, Perl, Ruby]
...

Nodes

Nodes are entities in the YAML informational model. There are three kinds of nodes: scalar, sequence, and mapping. In PyYAML, nodes are produced by Composer and can be serialized to a YAML stream by Serializer.

ScalarNode(tag, value, style, start_mark, end_mark)
SequenceNode(tag, value, flow_style, start_mark, end_mark)
MappingNode(tag, value, flow_style, start_mark, end_mark)

The style and flow_style flags have the same meaning as for events. The value of a scalar node must be a unicode string. The value of a sequence node is a list of nodes. The value of a mapping node is a dictionary which keys and values are nodes.

Example:

>>> print yaml.compose("""
... kinds:
... - scalar
... - sequence
... - mapping
... """)

MappingNode(tag=u'tag:yaml.org,2002:map', value={
    ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'kinds'): SequenceNode(tag=u'tag:yaml.org,2002:seq', value=[
        ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'scalar'),
        ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'sequence'),
        ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'mapping')])})

>>> print yaml.serialize(yaml.SequenceNode(tag=u'tag:yaml.org,2002:seq', value=[
...     yaml.ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'scalar'),
...     yaml.ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'sequence'),
...     yaml.ScalarNode(tag=u'tag:yaml.org,2002:str', value=u'mapping')]))

- scalar
- sequence
- mapping

Loader

Loader(stream)
SafeLoader(stream)
BaseLoader(stream)

Loader(stream) is the most common of the above classes and should be used in most cases. stream is an input YAML stream. It can be a string, a Unicode string, an open file, an open Unicode file.

Loader supports all predefined tags and may construct an arbitrary Python object. Therefore it is not safe to use Loader to load a document received from an untrusted source. By default, the functions scan, parse, compose, construct, and others use Loader.

SafeLoader(stream) supports only standard YAML tags and thus it does not construct class instances and probably safe to use with documents received from an untrusted source. The functions safe_load and safe_load_all use SafeLoader to parse a stream.

BaseLoader(stream) does not resolve or support any tags and construct only basic Python objects: lists, dictionaries and Unicode strings.

Loader.check_token(*TokenClasses)
Loader.peek_token()
Loader.get_token()

Loader.check_token(*TokenClasses) returns True if the next token in the stream is an instance of one of the given TokenClasses. Otherwise it returns False.

Loader.peek_token() returns the next token in the stream, but does not remove it from the internal token queue. The function returns None at the end of the stream.

Loader.get_token() returns the next token in the stream and removes it from the internal token queue. The function returns None at the end of the stream.

Loader.check_event(*EventClasses)
Loader.peek_event()
Loader.get_event()

Loader.check_event(*EventClasses) returns True if the next event in the stream is an instance of one of the given EventClasses. Otherwise it returns False.

Loader.peek_event() returns the next event in the stream, but does not remove it from the internal event queue. The function returns None at the end of the stream.

Loader.get_event() returns the next event in the stream and removes it from the internal event queue. The function returns None at the end of the stream.

Loader.check_node()
Loader.get_node()

Loader.check_node() returns True is there are more documents available in the stream. Otherwise it returns False.

Loader.get_node() construct the representation graph of the next document in the stream and returns its root node.

Loader.check_data()
Loader.get_data()

Loader.add_constructor(tag, constructor) # Loader.add_constructor is a class method.
Loader.add_multi_constructor(tag_prefix, multi_constructor) # Loader.add_multi_constructor is a class method.

Loader.construct_scalar(node)
Loader.construct_sequence(node)
Loader.construct_mapping(node)

Loader.check_data() returns True is there are more documents available in the stream. Otherwise it returns False.

Loader.get_data() constructs and returns a Python object corresponding to the next document in the stream.

Loader.add_constructor(tag, constructor): see add_constructor.

Loader.add_multi_constructor(tag_prefix, multi_constructor): see add_multi_constructor.

Loader.construct_scalar(node) checks that the given node is a scalar and returns its value. This function is intended to be used in constructors.

Loader.construct_sequence(node) checks that the given node is a sequence and returns a list of Python objects corresponding to the node items. This function is intended to be used in constructors.

Loader.construct_mapping(node) checks that the given node is a mapping and returns a dictionary of Python objects corresponding to the node keys and values. This function is intended to be used in constructors.

Loader.add_implicit_resolver(tag, regexp, first) # Loader.add_implicit_resolver is a class method.
Loader.add_path_resolver(tag, path, kind) # Loader.add_path_resolver is a class method.

Loader.add_implicit_resolver(tag, regexp, first): see add_implicit_resolver.

Loader.add_path_resolver(tag, path, kind): see add_path_resolver.

Dumper

Dumper(stream,
    default_style=None,
    default_flow_style=None,
    canonical=None,
    indent=None,
    width=None,
    allow_unicode=None,
    line_break=None,
    encoding=None,
    explicit_start=None,
    explicit_end=None,
    version=None,
    tags=None)
SafeDumper(stream, ...)
BaseDumper(stream, ...)

Dumper(stream) is the most common of the above classes and should be used in most cases. stream is an output YAML stream. It can be an open file or an open Unicode file.

Dumper supports all predefined tags and may represent an arbitrary Python object. Therefore it may produce a document that cannot be loaded by other YAML processors. By default, the functions emit, serialize, dump, and others use Dumper.

SafeDumper(stream) produces only standard YAML tags and thus cannot represent class instances and probably more compatible with other YAML processors. The functions safe_dump and safe_dump_all use SafeDumper to produce a YAML document.

BaseDumper(stream) does not support any tags and is useful only for subclassing.

Dumper.emit(event)

Dumper.emit(event) serializes the given event and write it to the output stream.

Dumper.open()
Dumper.serialize(node)
Dumper.close()

Dumper.open() emits StreamStartEvent.

Dumper.serialize(node) serializes the given representation graph into the output stream.

Dumper.close() emits StreamEndEvent.

Dumper.represent(data)

Dumper.add_representer(data_type, representer) # Dumper.add_representer is a class method.
Dumper.add_multi_representer(base_data_type, multi_representer) # Dumper.add_multi_representer is a class method.

Dumper.represent_scalar(tag, value, style=None)
Dumper.represent_sequence(tag, value, flow_style=None)
Dumper.represent_mapping(tag, value, flow_style=None)

Dumper.represent(data) serializes the given Python object to the output YAML stream.

Dumper.add_representer(data_type, representer): see add_representer.

Dumper.add_multi_representer(base_data_type, multi_representer): see add_multi_representer.

Dumper.represent_scalar(tag, value, style=None) returns a scalar node with the given tag, value, and style. This function is intended to be used in representers.

Dumper.represent_sequence(tag, sequence, flow_style=None) return a sequence node with the given tag and subnodes generated from the items of the given sequence.

Dumper.represent_mapping(tag, mapping, flow_style=None) return a mapping node with the given tag and subnodes generated from the keys and values of the given mapping.

Dumper.add_implicit_resolver(tag, regexp, first) # Dumper.add_implicit_resolver is a class method.
Dumper.add_path_resolver(tag, path, kind) # Dumper.add_path_resolver is a class method.

Dumper.add_implicit_resolver(tag, regexp, first): see add_implicit_resolver.

Dumper.add_path_resolver(tag, path, kind): see add_path_resolver.

YAMLObject

Deviations from the specification

need to update this section

  • rules for tabs in YAML are confusing. We are close, but not there yet. Perhaps both the spec and the parser should be fixed. Anyway, the best rule for tabs in YAML is to not use them at all.
  • Byte order mark. The initial BOM is stripped, but BOMs inside the stream are considered as parts of the content. It can be fixed, but it's not really important now.
  • Empty plain scalars are not allowed if alias or tag is specified. This is done to prevent anomalities like [ !tag, value], which can be interpreted both as [ !<!tag,> value ] and [ !<!tag> "", "value" ]. The spec should be fixed.
  • Indentation of flow collections. The spec requires them to be indented more then their block parent node. Unfortunately this rule many intuitively correct constructs invalid, for instance,
    block: {
    } # this is indentation violation according to the spec.
    
  • ':' is not allowed for plain scalars in the flow mode. ~{1:2} is interpreted as { 1 : 2 }.~

Notes

To do and long-term goals:

  • fix recursive objects, refactor Emitter, tabs, indentation for flow collections, indentation for scalars (min=1?), 'y' is !!bool,
  • libyaml3000 (my SoC proposal).

SpecBugs?